Purchase this article with an account.
A. H. Rashid, R. Channa, M. Shulman, E. Hatef, J. Talmud, M. Ibrahim, Z. Rentiya, A. Khwaja, D. Do, Q. Nguyen; Evaluation of the Reproducibility and Agreement of Optical Coherence Tomography Images for Patients With Neovascular Age Related Macular Degeneration. Invest. Ophthalmol. Vis. Sci. 2010;51(13):4397.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Optical coherence tomography (OCT) is currently employed to evaluate neovascular age-related macular degeneration (NVAMD). We compare intra-observer reproducibility and agreement for two Spectral-domain (SD) and one Time-domain (TD) OCT devices in patients with NVAMD. Thickness measurements before and after correction of boundaries on SpectralisTM images were also compared.
42 eyes with NVAMD were scanned using SpectralisTM and CirrusTM SD-OCT, and StratusTM TD-OCT. The same operator scanned each eye twice on the three devices. The intraclass correlation coefficient (ICC) was calculated for reproducibility; Bland-Altman (BA) plots helped determine agreement among machines. For SpectralisTM images, automated algorithms were manually corrected to follow the inner and outer boundaries. A paired t-test was used to compare the retinal thickness before and after correction. The percentage of images with a thickness difference of >48µm in the central 1mm diameter before and after correction was determined.
ICCs were greater than 0.95 for all three machines. Mean central 1mm foveal thickness in µm (+SD) was 239.53 + 82.63, 350.53 + 122.67, and 291.79 + 149.35 for StratusTM, SpectralisTM, and CirrusTM, respectively. BA plots gave a mean difference in µm (95% CI interval) of 43.3 (-52.6, 139.2), -50.3µm (-288.2, 187.3), and 94.8 (-30.9, 220.5) for CirrusTM vs. StratusTM, CirrusTM vs. SpectralisTM, and SpectralisTM vs. StratusTM, respectively. Thickness measurements before and after correction of boundaries were significantly different for all except the outer temporal subfield (p<0.05). 27.5% of the images of SpectralisTM had a difference of >48µm in thickness in the central 1mm sub-field after correction.
Statistically significant differences in thickness measurements before and after correction of boundaries on SpectralisTM images suggest that manual correction for accuracy is necessary. However, more than 70% of thickness differences were below 48µm, suggesting that differences may be within the clinically accepted variability of the device. ICC values indicate that readings are reliable when scanning a patient on the same machine, but large BA plot confidence intervals show that values from different machines do not agree well. We conclude that the same device gives reliable readings for a patient with NVAMD, but different machines may give very different thickness measurements.
This PDF is available to Subscribers Only